AdventHealth is betting OpenAI can buy clinicians back time and keep low-acuity cases out of hospitals, executives say. The Florida-based health system has deployed OpenAI models across clinician and patient workflows, calling those APIs from a microservices architecture that ingests clinical and administrative records, applies natural language processing, text generation and sentiment analysis, and surfaces summaries, personalised messages and recommendation-style outputs in clinician and patient interfaces. Executives say the aim is to return time to caregivers and to create a more structured front door to appropriate care. The rollout sits alongside multi-vendor projects, including a Biofourmis-supported Hospital at Home programme and a large imaging deployment with Aidoc.
AdventHealth has quietly layered OpenAI into the machinery of its whole-person care strategy. The health system connects to OpenAI models through a microservices stack that takes in both clinical and administrative records, runs natural language processing and text-generation tasks, and returns clinician-facing summaries, patient messages and recommendation-style outputs.
How the integration works
The technical design is straightforward and deliberate. API calls to pre-trained OpenAI models sit behind an application layer that exposes results through user interfaces for clinicians and care teams. The deployed tooling synthesises unstructured clinical notes, drafts personalised patient communications to reduce clerical workload, applies sentiment analysis to surface patient experience signals, and produces outputs intended to help clinicians and care teams consider appropriate pathways for care.
AdventHealth describes the environment as meeting standard healthcare controls. The system uses secure storage with encryption both in transit and at rest, and role-based access controls to limit who can see generated content. An internal advisory board reviews AI programmes for patient safety and privacy before deployment, and the OpenAI-enabled work sits inside a broader governance frame drawn from the system's participation in national initiatives.
Those initiatives are named. AdventHealth is a founding member of the Coalition for Health AI, known as CHAI, and of the Trustworthy, Responsible AI Network, or TRAIN. The system cites those groups as part of the responsible-use framework that guides where and how the models are applied.
Practical uses and partner projects
Rob Purinton, chief AI officer for AdventHealth, has framed the work around returning time to caregivers and improving patient access. "AI should support people across the organization, creating more time for human connection, sound clinical judgment and compassionate care," he said. Purinton has also described AI as a more structured "front door" to care, helping people understand symptoms and select the correct care setting.
The OpenAI integration isn't presented as a standalone solution. AdventHealth has contracted specialist vendors for discrete service lines and is integrating those offerings into the same operational picture. Biofourmis is supplying remote patient monitoring technology and an acute Hospital at Home platform that's designed to feed data into AdventHealth's Epic electronic health record. That integration aims to support continuous monitoring, arrhythmia and fall detection, and the scheduling of in-home ancillary services.
The immediate operational milestone is the launch of the Biofourmis-supported Hospital at Home programme at AdventHealth Winter Park, which the system plans to use as a template for scaling remote whole-person care across its network.
Separately, Aidoc was announced as a partner for a large-scale imaging AI deployment across dozens of AdventHealth hospitals, freestanding emergency departments and imaging centres in Florida and Kentucky. Aidoc's deployment is intended to triage urgent imaging findings and help prioritise cases. Aidoc's release notes also referenced recent regulatory and capital milestones for the vendor, including a Breakthrough Device designation and a $150 million funding round.
AdventHealth emphasises augmentation rather than replacement. Victor Herrera, MD, chief clinical officer for AdventHealth's Central Florida Division, said the system values "the opportunity to share our knowledge and experience in using artificial intelligence to provide whole-person care." In practice that means applying model outputs to reduce documentation time, speed chart review and ease administrative preparation so clinicians can spend more time with patients.
What the brief technical account calls recommendation-style outputs are intended to aid triage and care-setting decisions that keep lower-acuity care outside the hospital. That's a practical, cost-containment aim as much as a patient experience one: routing callers to the appropriate level of care can reduce avoidable emergency department visits and free up inpatient capacity for higher acuity cases.
AdventHealth hasn't disclosed system-wide rollout dates, regional pricing or per-product licensing terms for the OpenAI-enabled tooling. The OpenAI work is being deployed alongside the other AI initiatives described above rather than replacing them. The system's public framing keeps returning to two numbers: the time clinicians might regain, and the number of sites where partner deployments will run, notably the dozens cited in the Aidoc imaging rollout and the single-site start for the Hospital at Home programme at AdventHealth Winter Park.
From a technical and governance standpoint, the combination of secure microservices, an internal advisory board, and membership of CHAI and TRAIN is intended to offer a balance. It aims to pair rapid adoption of generative capabilities with review processes focused on patient safety and privacy, and with third-party vendor relationships that cover specialised needs such as remote monitoring and imaging triage.
The immediate practical tests will be those partner rollouts. Biofourmis's Hospital at Home work will test continuity of care between in-home monitoring and Epic, while Aidoc's imaging triage will be evaluated across multiple hospitals and imaging sites in two states. Success will be judged on whether clinicians experience measurable reductions in administrative work and whether patients receive more timely, appropriate care settings as a result.
Related Articles
- OpenAI says Musk warned Brockman and Altman would face hatred
- Anthropic and OpenAI launch rival enterprise joint ventures
- Small models beat large ones, 60% better accuracy
AdventHealth's next concrete milestone is clear: the system will use the Biofourmis-supported Hospital at Home launch at AdventHealth Winter Park as a template for scaling remote, whole-person care across its network. That event will be the first operational test of how OpenAI-enabled summaries, personalised messaging and recommendation outputs work in concert with vendor platforms and Epic to return time to clinicians and steer patients to the right setting.
This article was created with AI assistance.
